U.S. patent application number 12/421688 was filed with the patent office on 2009-08-06 for method and system for indexing information about entities with respect to hierarchies.
This patent application is currently assigned to Initiate Systems, Inc.. Invention is credited to James B. Cushman, II, Scott Ellard.
Application Number | 20090198686 12/421688 |
Document ID | / |
Family ID | 38750753 |
Filed Date | 2009-08-06 |
United States Patent
Application |
20090198686 |
Kind Code |
A1 |
Cushman, II; James B. ; et
al. |
August 6, 2009 |
Method and System for Indexing Information about Entities with
Respect to Hierarchies
Abstract
Systems and methods for indexing, associating or compositing
data records and hierarchies from various information sources are
disclosed. Embodiments of the present invention may provide the
ability to link data records and thus to link data records to known
hierarchies of data records. More specifically, embodiments of the
present invention may provide the capability to associate data
records in varying information sources and to thereby associate
incoming data record with existing data records or existing data
hierarchies such that an incoming data record may not only be
associated with an existing data record comprising information
about the same entity but may additionally be associated with other
members of the data hierarchy in the same manner as the existing
data record. In addition to associating an incoming data record
with an existing data record and incorporating the incoming data
record into an existing data hierarchy, embodiments of the present
invention may provide the capability of reconciling an incoming
data hierarchy to which an incoming data record belongs with an
existing data hierarchy belongs such that the two data hierarchies
may be composited.
Inventors: |
Cushman, II; James B.;
(Alpharetta, GA) ; Ellard; Scott; (Marietta,
GA) |
Correspondence
Address: |
SPRINKLE IP LAW GROUP
1301 W. 25TH STREET, SUITE 408
AUSTIN
TX
78705
US
|
Assignee: |
Initiate Systems, Inc.
Chicago
IL
|
Family ID: |
38750753 |
Appl. No.: |
12/421688 |
Filed: |
April 10, 2009 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
11656111 |
Jan 22, 2007 |
7526486 |
|
|
12421688 |
|
|
|
|
60802356 |
May 22, 2006 |
|
|
|
Current U.S.
Class: |
1/1 ;
707/999.005; 707/E17.017 |
Current CPC
Class: |
G06F 16/215
20190101 |
Class at
Publication: |
707/5 ;
707/E17.017 |
International
Class: |
G06F 17/30 20060101
G06F017/30 |
Claims
1. A method for associating data records, comprising: receiving a
data record; identifying a set of candidate data records based on a
comparison between a set of existing data records and the received
data record; scoring each of the set of candidate data records,
wherein the score of each of the candidate data records corresponds
to a likelihood that the first data record and the candidate data
record comprise information on an entity; and associating the
received data record with a first candidate record of the set of
candidate record if the score of the first candidate record is
greater than a first threshold, wherein the first candidate record
is in a first data hierarchy such that the first candidate data
record has a first set of hierarchical associations with a first
set of related data records and the received data record is
associated with the first candidate record such that the related
data record data has the first set of hierarchical associations
with the first set of related data records.
Description
RELATED APPLICATIONS
[0001] This application is a continuation of U.S. patent
application Ser. No. 11/656,111, entitled "Method and System for
Indexing Information about Entities with Respect to Hierarchies" by
inventors James B. Cushman II and Scott Ellard filed on Jan. 22,
2007, which claims a benefit of priority to the filing date of U.S.
Provisional Patent Application Ser. No. 60/802,356 by inventors
Scott Ellard and James B. Cushman II, entitled "System and Method
for Indexing Information about Entities to Hierarchies from
Different Information Sources" filed on May 22, 2006, the entire
contents of which are hereby expressly incorporated by reference
for all purposes.
TECHNICAL FIELD OF THE INVENTION
[0002] This invention relates generally to a system and method for
associating data records within one or more databases, and in
particular to a system and method for identifying data records in
one or more databases that may contain information about the same
entity and associating those data records together for easier
access to information about the entity. Even more particularly, the
present invention relates to associating one or more data records
in a hierarchy that may contain information about the same
entity.
BACKGROUND OF THE INVENTION
[0003] Data about entities, such as people, products, or parts may
be stored in digital format in a computer database. These computer
databases permit the data about an entity to be accessed rapidly
and permit the data to be cross-referenced to other relevant pieces
of data about the same entity. The databases also permit a person
to query the database to find data records pertaining to a
particular entity. The terms data set, data file, and data source
may also refer to a database. A database, however, has several
limitations which may limit the ability of a person to find the
correct data about an entity within the database. The actual data
within the database is only as accurate as the person who entered
the data. Thus, a mistake in the entry of the data into the
database may cause a person looking for data about an entity in the
database to miss some relevant data about the entity because, for
example, a last name of a person was misspelled. Another kind of
mistake involves creating a new separate record for an entity that
already has a record within the database. In a third problem,
several data records may contain information about the same entity,
but, for example, the names or identification numbers contained in
the two data records may be different so that the database may not
be able to associate the two data records to each other.
[0004] For a business that operates one or more databases
containing a large number of data records, the ability to locate
relevant information about a particular entity within and among the
respective databases is very important, but not easily obtained.
Once again, any mistake in the entry of data (including without
limitation the creation of more than one data record for the same
entity) at any information source may cause relevant data to be
missed when the data for a particular entity is searched for in the
database. In addition, in cases involving multiple information
sources, each of the information sources may have slightly
different data syntax or formats which may further complicate the
process of finding data among the databases. An example of the need
to properly identify an entity referred to in a data record and to
locate all data records relating to an entity in the health care
field is one in which a number of different hospitals associated
with a particular health care organization may have one or more
information sources containing information about their patient, and
a health care organization collects the information from each of
the hospitals into a master database. It is necessary to link data
records from all of the information sources pertaining to the same
patient to enable searching for information for a particular
patient in all of the hospital records.
[0005] There are several problems which limit the ability to find
all of the relevant data about an entity in such a database.
Multiple data records may exist for a particular entity as a result
of separate data records received from one or more information
sources, which leads to a problem that can be called data
fragmentation. In the case of data fragmentation, a query of the
master database may not retrieve all of the relevant information
about a particular entity. In addition, as described above, the
query may miss some relevant information about an entity due to a
typographical error made during data entry, which leads to the
problem of data inaccessibility. In addition, a large database may
contain data records which appear to be identical, such as a
plurality of records for people with the last name of Smith and the
first name of Jim. A query of the database will retrieve all of
these data records and a person who made the query to the database
may often choose, at random, one of the data records retrieved
which may be the wrong data record. The person may not often
typically attempt to determine which of the records is appropriate.
This can lead to the data records for the wrong entity being
retrieved even when the correct data records are available. These
problems limit the ability to locate the information for a
particular entity within the database.
[0006] To reduce the amount of data that must be reviewed and
prevent the, user from picking the wrong data record, it is also
desirable to identify and associate data records from the various
information sources that may contain information about the same
entity. There are conventional systems that locate duplicate data
records within a database and delete those duplicate data records,
but these systems only locate data records which are identical to
each other. Thus, these conventional systems cannot determine if
two data records, with for example slightly different last names,
nevertheless contain information about the same entity. In
addition, these conventional systems do not attempt to index data
records from a plurality of different information sources, locate
data records within the one or more information sources containing
information about the same entity, and link those data records
together.
[0007] These information sources may also impose hierarchical
relationships among the various data records pertaining to
different entities. These hierarchies may designate a variety of
relationships between entities, such as social hierarchies
(business organization, army chain of command, and church
organization), containment hierarchies (biological taxonomy,
geometric subsets, assemblies, bill of materials), genealogy
hierarchies, or other parent-child data relationships. Thus, not
only is it desirable to identify and associate data records from
various data sources, but it may also be desirable to associate
data records with a data records in an existing or known
hierarchy.
[0008] For example, a company may have multiple suppliers of parts
where the suppliers may belong to a hierarchy of parent companies
and there is a need to determine the level of business with a
particular parent company on an ongoing basis. Multiple information
sources may contain the different orders for parts from individual
companies, while another 3.sup.rd party source (such as Dunn &
Bradstreet, Equifax, infoUSA, etc.) identifies the parent company
hierarchy. It may be desirable to link part suppliers to the
hierarchy to determine the amount of business with any particular
parent company.
[0009] In addition to the problems discussed above with respect to
entity matching, the ability to match data records to known
hierarchies may present additional problems such as that there may
be missing parts of the hierarchy, a data record may match to more
than one node of a hierarchy tree, a data record may match to nodes
on two separate hierarchy trees or a data record which is a node on
one hierarchy tree may match to a node on another hierarchy tree
and thus it may be necessary to reconcile the two hierarchy trees
with one another.
[0010] Thus there is a need for a system and method for indexing
information about entities/hierarchies from a plurality of
different information sources which avoid these and other problems
of known systems and methods, and it is to this end that the
present invention is directed.
SUMMARY OF THE INVENTION
[0011] Systems and methods for indexing, associating or compositing
data records and hierarchies from various information sources are
disclosed. Embodiments of the present invention may provide the
ability to link data records and thus to link data records to known
hierarchies of data records. More specifically, embodiments of the
present invention may provide the capability to associate data
records in varying information sources and to thereby associate
incoming data record with existing data records or existing data
hierarchies such that an incoming data record may not only be
associated with an existing data record comprising information
about the same entity but may additionally be associated with other
members of the data hierarchy in the same manner as the existing
data record. In addition to associating an incoming data record
with an existing data record and incorporating the incoming data
record into an existing data hierarchy, embodiments of the present
invention may provide the capability of reconciling an incoming
data hierarchy to which an incoming data record belongs with an
existing data hierarchy belongs such that the two data hierarchies
may be composited.
[0012] In certain embodiments, the present invention may link data
records containing information about the same entity, to integrate
data records into existing data hierarchies and to composite (e.g.
join or merge) data hierarchies. In one particular embodiment, a
data record may be compared to existing data to locate data records
containing information about the same entity. The matching
operation may use one or more combinations of attributes to
retrieve a plurality of candidates, generate a confidence level or
score for each candidate and identify data records which have
scores greater than or equal to a threshold level. The data record
may then be associated with one of the identified data records,
integrated into a data hierarchy to which the identified data
record belongs, or a data hierarchy to which the data record
belongs composited with a data hierarchy to which the identified
data record belongs.
[0013] Embodiments of the present invention may provide the
technical advantages that data record from various information
sources may be integrated into existing data hierarchies based on a
statistical algorithms, resulting in the disambiguation of various
data records and data hierarchies which may be received from a
variety of different sources.
[0014] Embodiments of the invention disclosed herein can be
implemented by programming one or more computer systems or devices
with computer-executable instructions embodied in a
computer-readable medium. When executed by a processor, these
instructions operate to cause these computer systems and devices to
perform one or more functions particular to embodiments of the
invention disclosed herein (e.g., generate an appropriate
confidence level or score for each event.) Programming techniques,
computer languages, devices, and computer-readable media necessary
to accomplish this are known in the art and thus will not be
further described herein.
[0015] These, and other, aspects of the invention will be better
appreciated and understood when considered in conjunction with the
following description and the accompanying drawings. The following
description, while indicating various embodiments of the invention
and numerous specific details thereof, is given by way of
illustration and not of limitation. Many substitutions,
modifications, additions or rearrangements may be made within the
scope of the invention, and the invention includes all such
substitutions, modifications, additions or rearrangements.
BRIEF DESCRIPTION OF THE DRAWINGS
[0016] FIG. 1 is a block diagram illustrating a database system
that may include a master entity index system in accordance with
the invention;
[0017] FIG. 2 is a block diagram illustrating a master entity index
system and its associated databases in accordance with the
invention;
[0018] FIG. 3 is a block diagram illustrating more details of the
database that are associated with the master entity index;
[0019] FIG. 4 is a flowchart illustrating a plurality of input
operations that may be executed by the master entity index of FIG.
2;
[0020] FIG. 5 is a flowchart illustrating a plurality of query
operations that may be executed by the master entity index of FIG.
2;
[0021] FIG. 6 is a flowchart illustrating a plurality of monitor
operations that may be executed by the master entity index of FIG.
2 (where the plurality of operations is referred to as a whole as
"exception processing");
[0022] FIG. 7 is a flowchart illustrating a new data record
addition operation that may be executed by the master entity index
of FIG. 2;
[0023] FIG. 8 is a flowchart illustrating an existing data record
update operation that may be executed by the master entity index of
FIG. 2;
[0024] FIG. 9 is a flowchart illustrating the match/link operation
that may be executed by the master entity index of FIG. 2;
[0025] FIG. 10 is a flowchart illustrating an identity rule
operation that may be executed by the master entity index of FIG.
2;
[0026] FIG. 11 is a flowchart illustrating a non-identity rule
operation that may be executed by the master entity index of FIG.
2;
[0027] FIG. 12 is a flowchart illustrating a delete operation that
may be executed by the master entity index of FIG. 2;
[0028] FIG. 13 is a flowchart illustrating a data record retrieval
operation that may be executed by the master entity index of FIG.
2;
[0029] FIG. 14 is a flowchart illustrating a database retrieval
operation that may be executed by the master entity index of FIG.
2;
[0030] FIG. 15 is a flowchart illustrating a match operation that
may be executed by the master entity index of FIG. 2;
[0031] FIGS. 16A and 168 are block diagrams illustrating more
details of embodiments of a database associated with a master
entity index;
[0032] FIGS. 17A, 17B, 17C and 18 are graphical representation of
example hierarchies of data records;
[0033] FIG. 19 is a flow diagram of one embodiment of a method for
associating data records;
[0034] FIG. 20 is a graphical representation of example hierarchies
of data records;
[0035] FIGS. 21-26 are graphical representations of examples of
associating a data record with a hierarchy of data records;
[0036] FIGS. 27A and 27B are flow diagrams of one embodiment of a
method for associating data records with data hierarchies; and
[0037] FIGS. 28 and 29 are graphical representations of examples of
associating a data record with a hierarchy of data records.
DETAILED DESCRIPTION
[0038] The invention and the various features and advantageous
details thereof are explained more fully with reference to the
nonlimiting embodiments that are illustrated in the accompanying
drawings and detailed in the following description. Descriptions of
well known starting materials, processing techniques, components
and equipment are omitted so as not to unnecessarily obscure the
invention in detail. Skilled artisans should understand, however,
that the detailed description and the specific examples, while
disclosing preferred embodiments of the invention, are given by way
of illustration only and not by way of limitation. Various
substitutions, modifications, additions or rearrangements within
the scope of the underlying inventive concept(s) will become
apparent to those skilled in the art after reading this
disclosure.
[0039] Reference is now made in detail to the exemplary embodiments
of the invention, examples of which are illustrated in the
accompanying drawings. Wherever possible, the same reference
numbers will be used throughout the drawings to refer to the same
or like parts (elements). In one embodiment, the system and method
of the invention is particularly applicable to a system and method
for indexing information from multiple information sources about
companies to an explicit business hierarchy such as Dun and
Bradstreet (D&B), Experian, or Equifax. It is in this context
that the invention will be described. It will be appreciated,
however, that the system and method in accordance with the
invention has utility in a large number of applications that
involve identifying, associating, and structuring into hierarchy
information about entities.
[0040] In describing embodiments of the systems and methods of the
present invention, it may first be helpful to go over examples of
embodiments of systems and methods for associating entities which
may be utilized in conjunction with embodiments of the present
invention such has those described in U.S. Pat. No. 5,991,758,
entitled "System and Method for Indexing Information about Entities
from Different Information Sources", issued Nov. 23, 1999 by
inventor Scott Ellard hereby incorporated by reference in its
entirety. FIG. 1 is a block diagram illustrating a master entity
index system 30 in accordance with the invention. The master entity
index system may include a master entity index (MEI) 32 that
processes, updates and stores data records about one or more
entities from one or more information sources 34, 36, 38 and
responds to commands or queries from a plurality of operators 40,
42, 44, where the operators may be either users or information
systems. The MEI may operate with data records from a single
information source or, as shown, data records from one or more
information sources. The entities tracked using the MEI may include
for example, patients in a hospital, participants in a health care
system, parts in a warehouse or any other entity that may have data
records and information contained in data records associated with
it. The MEI may be a computer system with a central processing unit
45 executing a software application that performs the function of
the MEI. The MEI may also be implemented using hardware
circuitry.
[0041] As shown, the MEI 32 may receive data records from the
information sources as well as write corrected data back into the
information sources. The corrected data communicated to the
information sources may include information that was correct, but
has changed, information about fixing information in a data record
or information about links between data records. In addition, one
of the users 40-44 may transmit a query to the MEI 32 and receive a
response to the query back from the MEI. The one or more
information sources may be, for example, different databases that
possibly have data records about the same entities. For example, in
the health care field, each information source may be associated
with a particular hospital in the health care organization and the
health care organization may use the master entity index system to
relate the data records within the plurality of hospitals so that a
data record for a patient in Los Angeles may be located when that
same patient is on vacation and enters a hospital in New York. The
MEI 32 of the master entity index system 30 may be located at a
central location and the information sources and users may be
located remotely from the MEI and may be connected to the MEI by,
for example, a communications link, such as the Internet. The MEI,
the one or more information sources and the plurality of users may
also be connected together by a communications network, such as a
wide area network. The MEI may have its own database that stores
the complete data records in the MEI, but the MEI may also only
contain sufficient data to identify a data record (e.g., an address
in a particular information source) or any portion of the data
fields that comprise a complete data record so that the MEI
retrieves the entire data record from the information source when
needed. The MEI may link data records together containing
information about the same entity in an entity identifier or
associative database, as described below, separate from the actual
data record. Thus, the MEI may maintain links between data records
in one or more information sources, but does not necessarily
maintain a single uniform data record for an entity. Now, an
example of the master entity index system for a health care
organization in accordance with the invention will be
described.
[0042] FIG. 2 is a block diagram illustrating an example of a
master entity index system 50 for a health care organization. In
this example, the master entity index system may include a master
entity index 52 and a data store 54. For clarity, the one or more
information sources and the multiple users are not shown, but are
connected to the master entity index 52 as previously described.
The data store 54 may include an entity database 56, one or more
control databases 58, and an exception occurrence database. The
entity database may store the data from the data records as
specified above from the one or more information sources and may
separately store links between one or more data records when those
data records contain information about the same entity. The entity
database may also store an address of a large data record stored in
one of the information sources to reduce the storage requirements
of the entity database. In this example, the information about
entities within the data records may be information about patients
within a plurality of hospitals which are owned by a health care
organization. The MEI 52 may process the data records from the one
or more information sources located at each hospital, identify and
associate records that contain information about the same entity,
and generate the links between the separate data records when the
data records contain information about the same patient.
[0043] As data records from the information sources are fed into
the MEI, the MEI may attempt to match the incoming data record
about an entity to a data record already located in the MEI
database. The matching method will be described below with
reference to FIG. 15. If the incoming data record matches an
existing data record, a link between the incoming data record and
the matching data record may be generated. If the incoming data
record does not match any of the existing data records in the MEI,
a new entity identifier, as described below, may be generated for
the incoming data record. In both cases, the incoming data record
may be stored in the MEI. Then as additional data records are
received from the information sources, these data records are
matched to existing data records and the MEI database of data
records is increased.
[0044] The one or more control databases 58 may be used by the MEI
to control the processing of the data records to increase accuracy.
For example, one of the control databases may store rules which may
be used to override certain anticipated erroneous conclusions that
may normally be generated by the MEI. For example, the operator of
the MEI may know, due to past experience, that the name of a
particular patient is always misspelled in a certain way and
provide a rule to force the MEI to associate data records with the
known different spellings. The control databases permit the
operator to customize the MEI for a particular application or a
particular type of information. Thus, For a health care system
containing information about a patient, the control databases may
contain a rule that the nickname "Bill" is the same as the full
name "William." Therefore, the MEI will determine that data records
otherwise identical except for the first name of "Bill" and
"William" contain information about the same entity and should be
linked together. The MEI will now be described in more detail.
[0045] FIG. 3 is a block diagram illustrating more details of the
master entity index system 50, and in particular the MEI 52 and the
data store 54. The MEI 52 may include an addition and updating unit
70, a monitor unit 72 and a query unit 74. The addition and
updating unit may add data records about a new entity into the data
store, update data records in the data store, or add new rules to
the control databases. The monitor unit may permit a user of the
master entity index system to view special conditions, known as
exceptions, generated by the MEI. For example, a data record that
requires a person to view the data record due to an error may be
tagged and a message to the operator may be generated. The query
unit permits a user of the master entity index system to query the
MEI about information in the data records or information in the
control databases of the MEI and the MEI will return a response to
the query including any relevant data records or information. More
details of these units and their associated functions will be
described below.
[0046] For each of the operations of the MEI, including the
synthesis, as described below, the querying and the monitoring, the
results of those operations may depend on a trust value that may be
associated with each data field in a data record. The trust
computation for a data field may vary depending on the
characteristics of the data field, such as the date on which that
data record containing the field was received, or a quantitative
characterization of a level of trust of the information source. For
example, a data field containing data that was manually entered may
have a lower trust value than a data field with data that was
transferred directly from another information source. The trust
value for a data field may also affect the probability of the
matching of data records. Now, the data store 54 of the master
entity index system will be described in more detail.
[0047] The MEI may provide other operations that can be constructed
from combining the operations listed above. For example, an
operation to process data records for which it is not known if a
data record exists can be constructed by combining the query
operation for data records with the add new data record or update
existing data record operations. These "composite" operations may
lead to better performance than if the operator executed a
combination of the basic operations. They also relieve the operator
for having to determine the correct sequencing of operations to
achieve the desired result.
[0048] The data store 54 may include an entity database 56, one or
more control databases 58, and an exception occurrence database 90
as described above. The entity database may include a data record
database 76 and an identity database 78. The data record database
may store the data records or the addresses of the data records in
the MEI, as described above, while the associative identity
database may store a group of data record identifiers that
associate or "link" those data records which contain information
about the same entity. The separation of the physical data records
from the links between the data records permits more flexibility
because a duplicate copy of the data contained in the data record
is not required to be present in the identity database. The data
record database and the associative database may also be combined
if desired.
[0049] The identity database represents the combination of data
records in the data record database that refer to the same entity.
Each entity is assigned an entity identifier. Entity identifiers
are based on the concept of "versioned" identification. An entity
identifier consists of a base part and a version number. The base
part represents a specific individual about whom information is
being linked. The version number represents a specific combination
of data records that provides information about the entity that is
known at a specific time. In this example, the data records are
shown as squares with the alphabetic identifier of the data record
inside, and the entity identifier is shown as the base part
followed by a period followed by a version number. For example,
"100.0" indicates an entity identifier with 100 as the base part
and 1 as the version number. In this example, entity identifier
100.0 links data records A and B, entity identifier 101.0 links
data records C, D and E, and entity identifier 101.1 links data
records A, B, and R. Now, the details of the control databases will
be described.
[0050] The one or more control databases 58 may permit the operator
of the master entity index system to customize the MEI's processing
based on information known to the operator. The control databases
shown are merely illustrative and the MEI may have additional
control databases which further permit control of the MEI by the
operator. The control databases may, for example, include a rules
database 80, an exception handling database 82, an anonymous name
database 84, a canonical name database 86, and a thresholds
database 88.
[0051] The rules database may contain links that the operator of
the system has determined are certain and should override the logic
of the matching of the MEI. For example, the rules database may
contain identity rules (i.e., rules which establish that a link
exists between two data records) and/or non-identity rules (i.e.,
rules which establish that no link exists between two data
records). In this example, the rules database contains identity
rules which are A=B and C=D and a non-identity rule which is
Q.notequal.R. These rules force the MEI to establish links between
data records or prevent links from being established between data
records. For example, the information sources may have four
patients, with data records S, T, U, and V respectively, who are
all named George Smith and the operator may enter the following
nonidentity rules (i.e. S.notequal.T, T.notequal.U, U.notequal.V,
V.notequal.S) to keep the data records of the four different
entities separate and unlinked by the MEI. The rules in the rules
database may be updated, added or deleted by the operator of the
master entity index system as needed.
[0052] The exception handling database 82 contains one or more
exception handling routines that permit the master entity index
system to handle data record problems.
[0053] The exception handling rules within the database may have
the form of "condition.fwdarw.action" processing rules. The actions
of these rules may be actions that the MEI should automatically
take in response to a condition, for example, to request that an
individual manually review a data record. An example of an
exception handling rule may be, "if duplicate data
record.fwdarrow.delete data record" which instructs the MEI to
delete a duplicate data record. Another example is, "if different
attributes (sex).forwardarrrow.request further review of data
record" which instructs the MEI that if there are two data records
that appear to relate to the same entity, but the sex of the entity
is different for each data record, the MEI should request further
review of the data records. In response to this request, an
operator may determine that the data records are the same, with a
incorrectly typed sex for one of the records and the operator may
enter a rule into the rules database that the two data records are
linked together despite the difference in the sex attribute. The
exception database may have an associated database 80 (described
below) which stores the actual exceptions that occur during
processing of the input data records.
[0054] The anonymous name database 84 permits the MEI to
automatically recognize names that should be ignored for purposes
of attempting to match two data records. In this example, the
anonymous name database may contain "not on file", "john doe" and
"baby.subtext.--1" which are names that may be typically assigned
by a hospital to a patient when the hospital has not yet determined
the name of the patient. As another example, a part not in a
warehouse inventory may be referred to as "not on file" until the
part may be entered into the database. These anonymous names may be
used by the MEI to detect any of the anonymous names or other
"filler" data that hold a space, but have no particular meaning in
data records and ignore those names when any matching is conducted
because a plurality of data records containing the name of "john
doe" should not be linked together simply because they have the
same name.
[0055] The canonical name database 86 may permit the MEI to
associate short-cut data, such as a nickname, with the full data
represented by the short-cut data, such as a person's proper name.
In this example for a health care organization, the nickname Bill
may be associated with William and Fred may be associated with
Frederick. This database permits the MEI to link together two data
records that are identical except that one data record uses the
first name Bill while the second data record uses the first name
William. Without this canonical name database, the MEI may not link
these two data records together and some of the information about
that patient will be lost. The thresholds database 88 permits the
thresholds used by the MEI for matching data records, as described
below, to be adjustable. For example, an operator may set a high
threshold so that only exact data records are matched to each
other. A lower threshold may be set so that a data record with
fewer matching data fields may be returned to the user in response
to a query. The details of the matching method will be described
below in more detail.
[0056] The exception occurrence database 80 allows the MEI to
maintain a record of all of the exceptions that have occurred. The
exception occurrence database may store the actual exception
conditions that have arisen during processing. For example, the
exception occurrence database may contain an entry that represents
that entity 100.2 has two data records with different values for
the "sex" attribute.
[0057] The operator of the MEI may clear the identity database 78
without clearing the data record database 80. Thus, an operator may
have the MEI receive a plurality of input data records and generate
a plurality of links with a particular matching threshold level, as
described below, being used. The operator may then decide to
perform a second run through the data using a lower matching
threshold level to produce more links, but does not want to delete
the data records themselves, and does not want to delete the
identity and non-identity rules from the rules database created
during the first run through the data. Thus, the operator may
delete the identity database, but keep the control databases, and
in particular the rules database, for the second run through the
data. Now, a method of adding or updating data in the master entity
index in accordance with the invention will be described.
[0058] FIG. 4 is a flowchart illustrating a method 100 for adding
or updating data within the master entity index system. The user
selects an add/update operation in step 102 which permits the user
to select, for example, an add new data record operation 104, an
update an existing data record operation 106, an add new identity
rule 110, an add new non-identity rule 112, and a delete data
record operation 113. The add new data record operation permits a
user of the MEI to add a new data record containing information
about an entity into the MEI while the update an existing data
record operation permits a user of the system to update the data
record or information about an entity that already exists within
the MEI. The add identity and add non-identity rule operations
permit the user to add identity or nonidentity rules into the rules
database 80 shown in FIG. 3. The delete operation permits the user
of the MEI to delete a data record from the data records database.
Each of these operations will be described in more detail below
with reference to FIGS. 7-12. The MEI may then determine whether
there are additional addition or updating operations to perform in
step 114 based on the user's response and either exit the method or
return to step 102 so that the user may select another addition or
updating operation. The add/update/delete operation may also be
used for the control databases to add/update information in those
databases, and additional processing may occur due to changes in
the control databases which may change the identity database. In
all of those cases, the additional processing is to identify the
existing identity records that are impacted by the modification,
and to use the match/link operation to re-compute the appropriate
entries in the identity database. For example, removing a record
for the anonymous name database would cause re-computation of
identities of all records with that anonymous name, and all records
linked to those records.
[0059] For all of the data records stored by the MEI, a record
identifier may be used to uniquely identify the entity referred to
by that record compared to other data records received from the
data source. For example, in data records obtained from a hospital
information system, an internally-generated patient identifier may
be used as a record identifier, while in data records from a health
plan membership database, a social security number can be used as a
record identifier. A record identifier differs from an entity
identifier because its scope is only the data records from a single
data source. For example, if a person in a health plan is a patient
in the hospital, their hospital record will have a different record
identifier than their health plan record. Furthermore, if records
from those two data sources happened to have the same record
identifier, this would be no indication that the records referred
to the same entity.
[0060] An additional aspect of the data record database is that one
or more timestamps may be recorded along with the data record. The
timestamps may indicate when the data record was last changed
(e.g., when the data record is valid) and when the data record was
received from the information source. The timestamps may be used to
track changes in a data record which may indicate problems, such as
fraud, to the operation of the MEI. The timestamps may be generated
whenever a data record is added to the MEI or updated so that the
historical changes in the data record may be documented.
Additionally, individual attribute values may be associated with
status descriptors that describe how the values should be used. For
example, an attribute value with an "active" status would be used
for identification, an attribute value with an "active/incorrect"
status would be used for identification but not presented to the
operator as being the correct value (for example, an old address
that still occurs in some incoming data records), and a status of
inactive/incorrect means that the value should no longer be used
for matching but should be maintained to facilitate manual review.
Now, a method for querying the MEI in accordance with the invention
will be described.
[0061] FIG. 5 is a flowchart illustrating a method 120 for querying
the master entity index in accordance with the invention. The
querying operations permit the user to retrieve information from
the MEI about a particular entity or data from one of the control
databases. After a user selects the query operation in step 122,
the user may select from a particular query operation that may
include an entity retrieval operation 124, or a database query
operation 128. For the entity retrieval operation, the MEI in step
132 may execute the match operation 300 described below. During the
match operation, an input query may be matched against data records
within the various information sources, as described in more detail
below with reference to FIG. 15. For the database retrieval
operation, the operator specifies a database and a set of attribute
values that indicates the records of interest. The MEI in step 136
may locate those records in the specified database that has
corresponding values for the specified attributes.
[0062] Additional queries may be performed by the MEI. The MEI may
be queried about the number of entities in the MEI database and the
MEI may respond with the number of entities in the MEI database.
The MEI may also be queried about the volatility (e.g., the
frequency that the data records change) of the data in the data
records using a timestamp indicating the last time and number of
times that the data has been changed that may be associated with
each data record in the MEI. The volatility of the data may
indicate fraud if the data about a particular entity is changing
frequently. The MEI may also be queried about the past history of
changes of the data in the data records so that, for example, the
past addresses for a particular entity may be displayed. Once the
queries or matches have been completed, the data is returned to the
user in step 138. The MEI may then determine whether there are
additional queries to be performed in step 140 and return to step
122 if additional queries are going to be conducted. If there are
no additional queries, the method ends. Now, an exception
processing method that may be executed by the MEI will be
described.
[0063] FIG. 6 is a flowchart of a method for processing exceptions
150 that may be executed by the MEI. The input is data describing
the occurrence of an exception, for example, an entity whose data
records indicate two different values for the entity's sex. In step
152, the exception given as input to the operation is recorded in
the exception occurrence database. In step 154, the MEI determines
if there is an exception handling rule within the exception
handling database 82 for handling the anomaly, as shown in FIG. 3
As described above, the exception handling database contains a
plurality of rules for handling various types of exceptions. If an
exception handling rule is in the exception handling database, in
step 156, the MEI may perform the exception handling routine in the
database. The routine may generate a message for the operator or
may process the data using another software program. A message may
be displayed to the user in step 158. If there was not an exception
handling routine in the exception handling database, then a message
is printed for the user in step 158. The message may require the
user to perform some action or may just notify the operator of the
action being taken by the MEI in response to an exception. After
the message is displayed, the exception handling method has been
completed. Now, the operations that may be performed by the MEI
during the addition and updating data method will be described.
[0064] FIG. 7 is a flowchart illustrating a method 170 for
inserting a new data record into the MEI in accordance with the
invention. The insertion of a new data record for a new entity
usually occurs when a particular information source has determined
that the new data record should not refer to the same entity as any
other data record previously generated by the information
source.
[0065] For inserting a new data record into the MEI, a record
containing the new data is received by the MEI from the user. The
MEI may then attempt to validate and standardize the fields in the
new data record.
[0066] Validation in step 172 may include examining the lengths of
the fields or the syntax or character format of the fields, for
example, as numeric fields may be required to contain digits in
specified formats. Validation may also involve validating codes in
the new data record, for example, valid state abbreviations or
diagnostic codes. Additional data sets may be involved in the
validation process, for example, a data set containing valid
customer account numbers. If the validation process fails, in step
176 an exception may be created that indicates that invalid data is
received, the exception handling method described above may be
performed, and processing of the insert new record operation is
complete.
[0067] During standardization in step 174, the MEI may process the
incoming data record to compute standard representations of certain
data items. For example, the incoming data record may contain the
first name of "Bill" and the MEI may add a matching field
containing "William" into the incoming data record so that the MEI
may match data records to William. This standardization prevents
the MEI from missing data records due to, for example, nicknames of
people. Other kinds of standardization may involve different coding
systems for medical procedures or standard representation of street
addresses and other geographic locations.
[0068] The MEI may then attempt in step 178 to determine if a data
record with the same record identifier already exists in the data
record database. If the standardized input data has the same record
identifier as an existing data record, in step 176 an exception may
be created that indicates that a two data records with the same
record identifier have been received, the exception handling method
described above may be performed, and processing of the insert new
record operation is complete. If the standardized input data does
not have the same record identifier as an existing data record,
then the standardized input data may be added into the MEI and a
timestamp may be added to the data record in step 180. Then in step
182, the match/link method 210 described below and summarized in
FIG. 15 may be performed. The match/link operation is initiated
using the standardized input data, and its execution makes the
results of the match/link operation available to the insert new
data record operation. Then in step 184, the MEI may determine if
the match/link operation linked the standardized input data record
with any other records from the same information source. If so, in
step 176 an exception may be created that indicates that a
duplicate data record has been received, the exception handling
method described above may be performed, and processing of the
insert new record operation is complete. If not, the results of the
match/link operation are returned to the operator and the insert
new data record operation has been completed. Now, a method for
updating an existing data record already in the MEI will be
described.
[0069] FIG. 8 is a flowchart illustrating a method 190 for updating
an existing data record containing information about a new or
existing entity in accordance with the invention. Updates occur
when an information source receives new information concerning an
entity for which is already in its data store. The new information
received by the information source will be communicated to the MEI
through the update operation.
[0070] To perform the update method, the MEI may first test the
input data for validity in step 191, using the same method as in
step 172 of the add new record operation described in FIG. 7. If
the validation process fails, in step 199 an exception may be
created that indicates that invalid data is received, the exception
handling method described above may be performed, and the
processing of the update existing data record operation is
complete. The MEI may then standardize the input data in step 192,
using the same method as in step 174 of the add new record
operation. The MEI may then attempt in step 193 to determine if a
data record with the same record identifier as the standardized
input data already exists in the data record database. If the
standardized input data does not have the same record identifier as
an existing data record, a new item may be added to the exception
database in step 199 indicating that a duplicate data record was
located, and no further processing is performed.
[0071] If the standardized input data does have the same record
identifier as an existing data record, then the incoming data
record is checked in step 193 to see if it contains exactly the
same values for data fields as a data record already contained in
the data record database. If the standardized input data does not
have the same record identifier as an existing data record, in step
199 an exception may be created that indicates that a duplicate
data record has been received, the exception handling method
described above may be performed, and processing of the update
existing data record operation is complete. If the standardized
input data contains exactly the same values, then the execution of
this operation cannot affect the identity database. As a result,
the timestamp of the existing data record may be updated in step
195 to reflect the current time and processing of the operation is
completed. If the standardized input data contains different field
values than the existing record with the same record identifier, in
step 196 the existing record's field values may be updated to be
consistent with the values in the standardized input data, and its
timestamp may be updated to reflect the current time. Since the
data in the existing record has now changed, the impact on the
identity database must be computed. To do this, the MEI in step 197
may first remove an entry in the identity database involving the
existing record, if such an entry exists. The MEI may then perform
a match/link operation in step 198 for the existing records and any
other records referred to in the identity database record removed
in step 197. These are the records that had been previously
recorded in the identity database as referring to the same entity
as the existing data record. The match/link operation performs as
described in FIG. 9.
[0072] Once the match/link results have been returned in step 198
or the timestamp updated in step 195 or an exception has been
generated in step 199, the add new data record operation has been
completed. Now, a method for matching/linking a data record will be
described.
[0073] FIG. 9 is a flowchart illustrating a method 210 for
matching/linking a data record in accordance with the invention.
This operation is used to determine the data records in the data
record database that refer to the same entity as an input data
record in the data record database.
[0074] To perform the match/link operation, in step 212, the MEI
may perform the match operation 300 described below and diagrammed
in FIG. 15. In this step, the data in the input data record is
given to the match operation as its input, and the data records
returned by the match operation are made available. The MEI may
then in step 214 determine if any matching data records were made
available. If no data records other than the input data record were
returned, the match/link operation is completed. If at least one,
other data record was returned, the incoming data record and
matching data records may be synthesized in step 216. The synthesis
process combines the data values in the new record and the existing
records associated with the entities. The MEI may then in step 218
determine if a condition indicating a synthesis exception has
occurred, as defined by the current contents of the exception
database. For example, if the incoming data record lists the sex of
the entity as male while one of the matching data records lists the
sex of the entity as female, and the exception database states that
coalescing records with different sexes is an exceptional
condition, an exceptional condition will be identified. If an
exception occurs, in step 220 the MEI may create and handle the
appropriate synthesis exception and the processing of the
match/link operation is complete. If there are no synthesis
exceptions, then in step 222, the MEI may determine the number of
identity records currently held in the identity database that link
data records which match the input data record. If no identity
records exist, in step 224, a record may be added to the identity
database with a new unique base part and a version number of 0. If
exactly one identity record exists, in step 226 the MEI may update
this record to add a link to the input data record. If more than
one identity record exists, the MEI in step 228 may "coalesce"
these records--that is, remove the existing identity records and
replaces them with a single identity record linking the input data
records with all the data records returned in step 212. After one
of steps 224, 226, and 228 are performed, the processing of the
match/link operation has been completed. Now, a method for adding
an identity rule in accordance with the invention will be
described.
[0075] FIG. 10 is a flowchart illustrating a method 240 for adding
an identity rule to the rules database of the MEI in accordance
with the invention. In step 242, the MEI may receive two data
record identifiers, I.subtext.1 and I.subtext.2. In this example,
the identity rule is I.subtext.1=I.subtext.2 which means that these
two data records contain information about the same entity. The MEI
may then determine if the two identifiers refer to separate unique
records in step 244 and an exception routine may be executed in
step 246 if an exception occurs. If there is no exception, the MEI
determines if the new identity rule is consistent with the rules
already contained in the rules database in step 248. If there is an
exception, such as the rules database has a non-identity rule that
specifies that I.subtext.1 and I.subtext.2 are not associated with
each other, an exception routine is executed in step 250. If the
new identity rule is consistent with the other rules in the rules
database, then the entity identifier containing the two data
records are synthesized in step 250 to determine if there are any
inconsistencies within the associations of the two entity
identifier as shown in step 252. If there are any inconsistencies
in the entity identifier, an exception handling routine is executed
in step 254. Otherwise, the entity identifier containing the two
data records are merged together in step 256 and the method is
completed. Now, a method of adding a non-identity rule to the rules
database in accordance with the invention will be described.
[0076] FIG. 11 is a flowchart illustrating a method 260 for adding
a non-identity rule to the rules database of the MEI in accordance
with the invention. In step 262, the MEI may receive two data
record identifiers, I.subtext.1 and I.subtext.2. In this example,
the non-identity rule is I.subtext.1.notequal.I.sub.2 which means
that these two data records contain information that is not about
the same entity. The MEI may then determine if the two identifiers
refer to separate unique records in step 264 and an exception
routine may be executed in step 266 if an exception occurs. If
there is no exception, the MEI determines if the new non-identity
rule is consistent with the rules already contained in the rules
database in step 268. If the new non-identity rule conflicts with
one of the existing rules in the rules database, an exception
occurs in step 270. If the new non-identify rule does not conflict,
then the MEI determines whether the two data records corresponding
to the identifiers are currently located in different entity
identifier in step 272. If the data records are already separated,
then the method ends. If the data records are not currently in
different entity identifiers, then in step 274 the data records
identified by I.subtext.1 and I.subtext.2 as well as the other data
records are removed from the entity identifier containing the data
records identified by I.subtext.1 and I.subtext.2 Then, in step
276, the match/link operation, as described above, is performed on
each data record removed from the entity identifier. The match/link
operation may re-associate those data records previously in the
entity identifier with other data records or reestablish the entity
identifier without either I.subtext.1 or I.subtext.2. Now, a method
for deleting data records in accordance with the invention will be
described.
[0077] FIG. 12 is a flowchart illustrating a method for deleting a
data record in accordance with the invention. In step 277, the MEI
determines if the data record to be deleted is located within an
entity identifier with other data records. If there are no other
data records in the entity identifier, then in step 278, the data
record may be deleted and the method is completed. If there are
other data records associated with the data record to be deleted,
then in step 279, all of the data records are removed from the
entity identifier, and in step 280, the selected data record may be
deleted. Then in step 281, a match/link operation, as described
above, is executed for the other data records previously in the
entity identifier. The match/link operation may re-associate those
data records previously in the entity identifier with other data
records or reestablish the entity identifier without the deleted
data records. Now, a method for querying the MEI for data records
and querying the MEI for information from the other control
databases will be described.
[0078] FIG. 13 is a flowchart illustrating a method 282 for
querying the MEI system for data records about a particular entity.
In step 283, the MEI accepts a query from the user that contains
entity attributes. These attributes correspond to data fields
within the data records stored by the MEI. In step 284, the MEI
retrieves data records which have data fields that match the
attributes provided in the query and displays those located data
records for the user. The details of the matching method will be
described below in method 300 and illustrated in FIG. 15.
[0079] FIG. 14 is a flowchart illustrating a method 290 for
querying the MEI to locate information in the databases of the MEI.
In step 292, the operator may input a database and values for
fields maintained in records of the database. In step 294, the MEI
may retrieve any information from the control databases relating to
the data record identifier I. For example, if the user queries the
MEI about rules in the rules database containing identifier I, the
MEI may return the identity rule I=M and the non-identity rule
I.notequal.N. Now, a method for computing the match operation data
records in the MEI database based on a set of query attributes will
now be described.
[0080] FIG. 15 is a flowchart illustrating a method 300 for finding
matching data records in the MEI database based on a set of query
attributes in accordance with the invention. In step 302, the MEI
accepts a query in the form of a list of entity attributes and
associated values. Examples of entity attributes in a health care
example could be patient number, first name, last name, or phone
number, or if the database is a parts inventory, the part number,
or the manufacturer for the part. In step 304, the threshold being
used by the matching operation may be retrieved from the thresholds
database shown in FIG. 3. As described above, the thresholds
database permits different threshold levels to be used depending on
how close a match is desired by the operator.
[0081] Once the threshold has been set, in step 306, a plurality of
candidates may be retrieved. To select the candidates, the input
attributes are divided into combinations of attributes, such as the
last name and phone number of the patient, the first name and last
name of a patient, and the first name and phone number of the
patient. The data records in the MEI database are exactly matched
against each combination of attributes to generate a plurality of
candidate data records. Determining candidates from several
combinations of attributes permits more fault tolerance because a
data record may have a misspelled last name, but will still be a
candidate because the combination of the first name and the phone
number will locate the data record. Thus, a misspelling of one
attribute will not prevent the data record from being a candidate.
Once the group of candidates has been determined, the confidence
level for each candidate data record may be calculated.
[0082] The confidence level may be calculated based on a scoring
routine, which may use historical data about a particular
attribute, such as a last address. Thus, if the current address and
past addresses match a query, the confidence level is higher than
that for a data record with the same current address but a
different old address. The scoring routine may also give a higher
confidence level to information more likely to indicate the same
entity, such as a social security number. The scoring routine may
add the confidence level for each attribute to generate a
confidence level value for a candidate record (match score). Once
the confidence levels have been calculated, any data records with
confidence levels higher than the threshold level are displayed for
the user in step 310. The method of matching attributes to data
records within the MEI database has been completed.
[0083] As mentioned above, data records may also be in hierarchical
relationships with one another. These hierarchical relationship may
or may not be determined by the MEI system 30 and may be provided
by information sources 34, 36, 38 and denote the relationships
between data records provided by that, or another, information
source 34, 36, 38. Examples of such information sources that
comprise data records and explicit hierarchical relationships among
those data records (e.g., parent-subsidiary corporations, etc.) are
Dun and Bradstreet, Experian, Acxiom, InfoUSA, etc. Alternatively,
these hierarchical relationships may be asserted explicitly (e.g.,
defined) between two or more data records in the MEI database
either automatically or through user input from an operator 40, 42,
44 for almost any reason, such as specialized knowledge, processing
by a sales territory management billing application, etc.
[0084] In any event, it is desirable to associate incoming data
records from an information source with existing data records and
integrating these incoming data records with existing data
hierarchies to which the existing data records belong. Incoming
data records may also belong to incoming data hierarchies (e.g.
data hierarchies specified by external sources), therefore it is
also desirable to match the incoming data records with existing
data records and reconcile the existing data hierarchies to which
the existing data records belong with the incoming data hierarchies
to which the incoming data records belong.
[0085] To that end, attention is now directed to systems and
methods for indexing, associating or compositing data records and
hierarchies from various information sources. Embodiments of the
present invention may provide the ability to link data records and
thus to link data records to known hierarchies of data records.
More specifically, embodiments of the present invention may provide
the capability to link data records in varying information sources
and to thereby link an incoming data record with existing data
records or existing data hierarchies such that an incoming data
record may be linked to an existing data record which comprising
information about the same entity (an identity link) or linked to
other members of the data hierarchy (referred to as hierarchy
links). In addition to identically linking an incoming data record
with an existing data record and incorporating the incoming data
record into an existing data hierarchy, embodiments of the present
invention may provide the capability of reconciling an incoming
data hierarchy to which an incoming data record belongs with an
existing data hierarchy belongs such that the two data hierarchies
may be composited.
[0086] Part and parcel with the above capabilities, embodiments of
the present invention may provide the ability to correctly and
properly identify an entity corresponding to a data record to
locate all data records relating to the entity or to locate all
data records hierarchically related to an entity. The master entity
index system may process incoming data records and compare them to
data records existing in the master entity index to locate data
records containing information about the same entity. The matching
operation may use one or more combinations of attributes to
retrieve a plurality of candidate data records, generate a
confidence level or match score for each candidate and only return
data records or associated hierarchy structures to the user which
have confidence levels greater than or equal to a configurable
threshold level or that have been specified as identical in a rule
database. The threshold level may be adjusted and the retrieval of
the candidates may use historical data about an entity during the
query. Based upon this confidence level, an incoming data record
may be associated with an existing data record (e.g., the two
records identity linked), the incoming data record may be linked
with an existing data hierarchy to which the existing data record
belongs (e.g., the incoming data record identically or
hierarchically linked with a data record in an existing data
hierarchy) or an incoming data hierarchy reconciled or composited
with an existing data hierarchy.
[0087] The one or more information sources may be, for example,
different databases that possibly have data records about the same
entities. For example, in the manufacturing industry, each
information source may be associated with different sub-assemblies,
and an external information source may provide a supplier business
hierarchy. The manufacturing organization may use the master entity
index system to relate purchased parts from vendors to each other
and to a business hierarchy. In this way, reports could be
generated detailing inventory of parts purchased from a particular
parent company. These information sources may be designated as
primary, secondary, tertiary, etc. such that data records or data
hierarchies received from various information sources may be
compared based on these precedence designations (i.e., in case of
disagreement, which source takes precedence over others).
[0088] To store data related to the data records and hierarchical
structures the entity database of the master entity index system
may comprise a link database for storing identity and hierarchy
links between data records. Thus, the storage of data records may
be separate from the storage of the links between the data records,
making the master entity index system more flexible. The one or
more control databases may permit the operator of the master entity
index to customize the operation of the master entity index or to
manually create and modify hierarchy structures.
[0089] Turning to FIG. 16A, a depiction of an embodiment of a
master entity index system 50 where the entity database 56 of MEI
52 includes identity/hierarchy database 78 operable to store both
identity links between data records and hierarchy links between
data records. The master entity index system 30 may link data
records containing information about the same entity (e.g., an
identity link) so that a search for that particular entity will
retrieve all the member data records that are linked together. The
master entity index system may also link data records in a variety
of data hierarchies (hierarchically link) so that a retrieval of a
particular data record or entity will retrieve all or a subset of a
data hierarchy (e.g. a set of entities, themselves comprising a set
of data records, where the data records or entities are directly or
indirectly (inferred) hierarchically linked to one another) to
which that data record or entity belongs, with or without
identically linked data records for each node in the hierarchy.
[0090] More particularly, in one embodiment, the identity/hierarchy
database 78 may store a group of data record identifiers that
associate or "link" those data records which contain information
about the same entity (identity link) and/or are hierarchically
related (hierarchy link). The identity/hierarchy database 78
represents the combination of data records in the data record
database that refer to the same entity, and/or belong to the same
hierarchy. Each entity is assigned an entity identifier, and has
link type of "X" meaning identity link or a link type of "P"
indicating a hierarchy link. Entity identifiers are based on the
concept of "versioned" identification. An entity identifier
consists of a base part and a version number. The base part
represents a specific individual about whom information is being
linked. The version number represents a specific combination of
data records that provides information about the entity that is
known at a specific time. In this example, the data records are
shown as letters, and the entity identifier is shown as the base
part followed by a period followed by a version number followed by
a link type of `X`. For example, "100.1:X" indicates an entity
identifier with 100 as the base part and 1 as the version number
and of X link type. Similarly, hierarchy information is shown as
the base part followed by a period followed by a version number
followed by a link type of "P". For example, "102.1:P indicates an
entity identifier with 102 as the base part, 1 as the version
number and a link type of P.
[0091] Referring specifically to FIG. 16A: entity identifier 99.1:X
links data records I and I (a self-identify link), entity
identifier 100.1:X links data records A and I, entity identifier
101.1:X links data records B and J, entity identifier 106.1:X links
data records C, D, E and K, "102.1: P" indicates data record G is
parented by data record H, "103.1:P" indicates data record H is
parented by data record I, "104.1:P" indicates data record J is
parented by data record H, and "105.1:P" indicates data record K is
parented by data record I.
[0092] Graphically, the data hierarchy represented by the explicit
hierarchy links of the above example may be depicted as a tree
structure, as shown in FIG. 17A. More particularly, link 1702
represents "102.1:P" indicating data record G 1712 parented by data
record H 1714, link 1704 represents "104.1:P" indicating data
record J 1716 parented by data record H 1714, link 1706 represents
"103.1:P" indicating data record H 1714 parented by data record I
1718 and link 1708 represents "105.1:P" indicating data record K
1720 parented by data record I 1716.
[0093] Though data records may not be explicitly hierarchically
linked in identity/hierarchy database 78 (e.g., linked by a P type
link), data records may be inferred hierarchically linked by virtue
of the fact that they are identically linked (e.g., X type link)
with one or more data records which are hierarchically linked. In
other words, when it has been determined that two data records
represent the same entity (e.g., matched or associated as described
above), every data record associated with that entity is deemed to
be in the same hierarchical relationship with other entities or
data records.
[0094] These concepts may be better explained with reference to the
example denoted in FIGS. 16A and 17A. Note that data record "A" is
not explicitly hierarchically linked to any other data record in
identity/hierarchy database 78, in other words there is no P type
link between data record "A" and any other data record in
identity/hierarchy database 78. Note additionally, however, that
data record "A" is identically related (X type link) to data record
"I" (e.g., data record "A" and "I" have been determined to
represent the same entity). Because of the identity link between
data record "A" and data record "I" (e.g., data record "I" has been
matched to data record "I"), data record "A" may be inferred
hierarchically related to the same data records to which data
record "I" is hierarchically related (both explicitly and
inferred). Thus, A may be inferred hierarchically related to data
records H and K (i.e., the parent of both) through its identity
link with data record "I".
[0095] In one embodiment, if an identity link is formed between a
first and a second data record, and a hierarchy link indicates that
the first data record is in the lower position of a data hierarchy
with respect to a third data record (e.g., the first data record is
parented by the third data record), the second data record cannot
be hierarchically linked to any other data records (e.g., the
second data record cannot be parented by any other data records as
it is transitively linked to the third data record through its
identity link with the second data record).
[0096] The above descriptions may be further elucidated upon with
reference to FIG. 18 which graphically depicts both the explicit
links of the example links contained in identity/hierarchy database
78 and the inferred hierarchy links that result from the identity
links of identity/hierarchy database 78. Node 1802 comprises entity
"100" comprising data record "I" and data record "A" (i.e.,
"100.1:X" linking data records A and I), node 1804 comprises entity
"H", node 1806 comprising entity 106 (i.e., "106.1:X" linking data
records C, D, E and K), node 1808 comprises entity 101 (i.e.,
"101.1:X" linking data records B and J"), and node 1810 comprises
data record "G".
[0097] Link 1822 represents "102.1:P" indicating data record G is
parented by data record H, link 1824 represents "104.1:P"
indicating data record J is parented by data record H, link 1826
represents "103.1:P" indicating data record H is parented by data
record I and link 1828 represents "105.1:P" indicating data record
K is parented by data record I. Note that because many of the data
records referenced by these links are associated with an entity
(e.g., linked with other data records); these explicit hierarchy
links imply a number of inferred hierarchy links. For example,
"105.1:P" indicating data record K is parented by data record I,
means that every data record associated with entity "106" to which
data record "K" belongs (represented by node 1806), namely data
records "C", "D", "E" and "K" is hierarchically linked to (parented
by) every data record belonging to entity "100" to which data
record "I" belongs, namely data records "A" and "I". Thought of
another way, the entities may be hierarchically related to one
another (e.g. members of each entity may be hierarchically linked
to one another) such that every data record which is a member of
one entity is hierarchically related in the same way to the data
records of the other entity.
[0098] Thus, if an incoming data record is linked to an existing
data record, that data record will be linked to the same entity as
the existing data record and thus be inferred hierarchically linked
to the same entities as those existing data records. Additionally,
if the existing data record has a hierarchy link where it is in the
lower position (e.g. parented by another data record/the child of
another data record) the incoming matching data record may not have
a hierarchy link formed where it is in the inferior position (e.g.,
may not be parented by any other data record). For example, data
records "C", "D" and "E" may not be hierarchically linked to any
other data record where the other data record is in a superior
position (e.g., data records "C", "D" and "E" may not be parented
by any other data records as they are inferred to be parented
thorough their respective identity links with data record "K").
[0099] Other embodiments of identity/hierarchy database 78 may
store identity links and hierarchy links in other manners.
Referring to FIG. 16B, another embodiment of a master entity index
system 50, where the entity database 56 of MEI 52 includes
identity/hierarchy database 78 operable to store both identity
links between data records and hierarchy links between entities, is
depicted. The master entity index system 50 may link data records
containing information about the same entity (e.g., an identity
link) so that a search for that particular entity will retrieve all
the member data records that are linked together. The master entity
index system 50 may also link entities in a variety of data
hierarchies (hierarchically link) so that a retrieval of a
particular entity will retrieve all or a subset of a data hierarchy
(e.g., a set of data records that are directly or inferred
hierarchically linked to one another) to which that entity
belongs.
[0100] More particularly, in one embodiment, the identity/hierarchy
database 78 may store a group of identifiers that associate or
"link" those data records which contain information about the same
entity (identity link) and/or are hierarchically related (hierarchy
link). In one embodiment a data hierarchy may comprise a set of
nodes associated with entities, one of the nodes being a root node,
where each node can have at most one parent and zero to many
children, each node (e.g., corresponding to an entity) may itself
be associated with zero or more data records. In this example, the
data records are represented as an alphabetic identifier and the
entity identifier is shown as the base part followed by a period
followed by a version number followed by a link type of "X". For
example, "100.1:X" indicates an entity identifier with 100 as the
base part and 1 as the version number and X link type. Similarly,
hierarchy information is shown as the base part followed by a
period followed by a version number followed by a link type of "P".
For example, "102.1:P" indicates an entity identifier with 102 as
the base part, 1 as the version number and a link type of P.
[0101] Referring specifically to FIG. 16B: entity identifier
"100.1:X" identity links data records I and A, entity identifier
"102.1:X" identity links data records K, C, D and E, entity
identifier "104.1:X" identity links data records B and J, entity
identifier "100.1:X:100" links entity 100 with itself (a
self-identity link designating a root node of a hierarchy),
"101.1:P" indicates node (entity) 101 is parented by node (entity)
100, "102.1:P" indicates node 102 is parented by node 100,
"1103.1:P" indicates node 103 is parented by node 101, and
"104.1:P" indicates that node 104 is parented by node 101.
[0102] Graphically, the data hierarchies represented by the
explicit hierarchy link of the above example may be depicted as a
tree structure shown in FIG. 17B. More particularly, link 1750
represents "103.1:P" indicating node 1752 corresponding to entity
103 (comprising data record G, not shown in FIG. 16B) parented by
node 1758 corresponding to entity 101, link 1754 represents
"104.1:P" indicating node 1756 corresponding to entity 104 (i.e.,
data records J and B represented by "104.1:X" link) parented by
node 101 1758, link 1760 represents "101.1:P" indicating node 1758
corresponding to entity 101 (e.g. comprising data record H)
parented by node 1764 corresponding to entity 100 (comprising data
records I and A) and link 1762 represents "102.1:P" indicating node
1764 corresponding to entity 102 parented by node 1764
corresponding to entity 100.
[0103] After the above discussion it may be realized that data
records may come from information sources 34, 36, 38 in a variety
of formats, may comprise a variety of different information
regarding an entity, etc. Thus, it may be desired to create a
standardized form of data record such that these data records may
comprise a uniform set of attributes in a uniform format which
correspond to an associated entity such that a user or operator can
manipulate or manage a data hierarchy, protect a data hierarchy
from change or alter relationships between entities without
altering data records from various external data sources. In one
embodiment, this standardized data record may be a master
organizational solution such as a standard out of the box customer
relationship management (CRM) solution or may be a proprietary
standard format.
[0104] By utilizing standard data record formats the state of a
data hierarchy at a particular time may be utilized to create a
master data record for each of the entities in the data hierarchy,
where the master data record for an entity may be a composite of
attributes of one or more of the set of data records associated
with the entity created using a set of rules for compositing the
data records, where the rules may take into account the precedence
level (e.g. primary, secondary, tertiary, etc.) of the source of
each of the set of data records. These master data records may be
maintained by MEI system 30 and linked with the entities from which
they were created. Thus, these master data records may similarly be
integrated into the data hierarchy utilized to create the master
data records. These master data records may be updated in
accordance with updating or alteration to the set of data records
from which they are created. For example, if an attribute of a data
record of the set of data records used to create the master data
record changes, the attribute may change with respect to the master
data record. The updating or changing of master data records may
also be accomplished in conjunction with a level of precedence of a
data source associated with a data record. For example, if a
changed data record is associated with a primary source the master
data record may be updated, while if it is from a secondary source
the master data record may not be updated.
[0105] By the same token, by creating master data records for each
of the entities in a data hierarchy a "snapshot" of a data
hierarchy may be created and recorded. That is, information of the
various data records associated with a data hierarchy at a
particular time in time may be captured in the master entity
records and these master entity records frozen or stored such that
the state of the data hierarchy at that particular time may be
accessed or referred to at a later time. Similarly, by freezing a
master data record changes to any data records of the set of data
records used to create the master data record (for example by the
information source from which they originate) may be ignored,
suspended, or promote follow-up review and resolution.
[0106] The creation of master data records may be better explained
with reference to FIG. 17C which depicts one embodiment of a master
data hierarchy created from the data hierarchy depicted in FIG.
17B. More particularly, master data record "1" 1772 is a composite
of data record "I" and data record "A" of node 1764 corresponding
to entity 100, master data record "2" 1774 is a composite of data
record "H" of node 1758 corresponding to entity 101 master data
record "3" 1776 is a composite of data record "K", data record "C",
data record "D" and data record "E" of node 1764 corresponding to
entity 102, master data record "4" 1778 is a composite of data
record "G" of node 1752 corresponding to entity 103 while master
data record "5" is a composite of data record "J" and data record
"B" of node 1756 corresponding to entity 104.
[0107] As discussed above, identity/hierarchy database 78 may be
populated through the processing of one or more external data
sources by MEI system 30, wherein the external data source may
designate a set of records and hierarchical information (e.g.,
hierarchy links) between the set of records. By processing the set
of records and hierarchical information from the external source
corresponding links representing relationships (e.g. identity and
hierarchy) between the set of records may be created. Relationships
in identity/hierarchy database 78 relationships may also be
asserted explicitly (e.g. defined) through user input from an
operator 40, 42, 44 for almost any reason, such as specialized
knowledge, processing by a billing application, etc. In any event,
identity/database 78 may comprise a set of existing data
hierarchies (e.g. data records associated with identity or
hierarchy links) such that if an coming data record is matched to
an existing data record, that data record will be linked to the
same entity as the existing data record and thus be hierarchically
linked to the same entities as the existing data records.
[0108] One embodiment for a method of matching incoming data
records with existing data records and linking incoming records
with existing data records (and thus may be linked with entities)
such that incoming data records are incorporated into an existing
data hierarchy is depicted in FIG. 19. In one embodiment, matching
can also occur across languages and locales (e.g. between data
records in different languages), if the incoming data record is
already associated with language/locale data records (associations
which may be provided by the source providing the data record), if
particular fields between data records are common (i.e. company
number), etc.
[0109] An incoming data record may be received at step 1910, after
which a set of candidate data records along with an associated
score (e.g. a confidence level or match score) for each of the
candidate data records may be generated at step 1920. For each of
the candidate data records, then, a category may be determined
based on the corresponding score at step 1930. In one embodiment,
there may be three categories: "Hard Link", "Soft Link", or "No
Link". A Hard Link denotes when a score for a candidate data record
is above a configurable automatic link threshold. In other words,
the two data records (e.g. the incoming data record and the
candidate data record) are considered by MEI system 30 to be same
entity. Soft Link denotes when a match score is below the automatic
link threshold and above a configurable review threshold, while No
Link denotes when a score is below the review threshold and thus
the data records are considered not the same entity.
[0110] If there are no candidate records with scores above the
configurable review threshold at step 1940, the incoming data
record may not be matched with any candidate records at step 1942
and may be assigned its own entity identifier and become the root
node of a separate data hierarchy. It can then be determined if
there are multiple candidate data records with scores above the
review threshold (e.g. that fall either into the Hard Link or Soft
Link category) at step 1950. If there is only one candidate data
record with a score above the review threshold (e.g. "No" branch of
step 1950), if the one candidate is above the automatic link
threshold at step 1960 (e.g. a Hard Link) an identity link may be
formed between the incoming data record and the candidate record at
step 1970, and thus the incoming data record may be identically
linked to the same entity (e.g. set of data records) to which the
candidate record is linked. If the candidate data record's score is
above the review threshold, the same type of linking may occur at
step 1980, however, this link may be tagged for later manual review
by an operator of MEI system 30. During this manual review, the
operator can make changes to the links stored in MEI system 30 as
needed.
[0111] Returning to step 1950, if there is more than one candidate
record with scores above the configurable review threshold ("Yes"
branch), the incoming data record may be linked with the entity of
the candidate record with the highest matching score at step 1990
as described above, however, this link may be tagged for later
manual review by an operator of MEI system 30 at step 1992. In one
embodiment, if two or more candidate data records have the same
match score the incoming data record may be linked to the candidate
data record associated with the lowest number entity identifier
(though any other methodology of selecting between candidate data
records with identical score may likewise be utilized).
[0112] The above discussed methodology may be better understood
with reference to the graphically depicted example of a data
hierarchy of FIG. 20. In FIG. 20, member data records A, B, C, D
and E are all hard linked to entities (e.g. one or more data
records associated with an entity). More specifically, data record
A 2002 may be associated with node 2004 representing an entity with
which data record I 2008 is associated. Data records C, D and E
2010, 2012, 2014 may be associated with node 2016 representing an
entity with which data record K 2018 is associated, etc. Incoming
data record data record X 2022 may have a Soft Link to data record
H 2024 associated with node 2028, and thus will be linked with data
record H 2024 and tagged and queued for manual review. Data record
Y 2026 may have either a Hard or Soft Link to data record H 2024
and data record G 2020 associated with node 2030, and thus will be
linked with the higher score (between data record H 2024 and data
record G 2020 when compared with data record Y 2026). Data record Z
2032 can have either a Hard or Soft Link to data records I 2008 and
data record M 2034, where data records I 2008 and M 2034 are
members of different data hierarchies. Data record Z 2032 will
again be linked with the higher match score (between data record I
2008 and data record M 2034 when compared with data record Z 3032)
and tagged and queued for manual review.
[0113] Note that when a data record is identically linked to
another data record (and thus is linked to an entity), the data
record is also inferred hierarchically linked to all the data
records which the data record to which it has been identically
linked is hierarchically linked (either explicitly or inferred).
For example, suppose an identical link is made between data record
Z 2032 and data record M 2034. Though no explicit hierarchy links
have been formed between data record Z 2032 and data records L and
N 2036, 2038, by virtue of the identity link formed between data
record Z 2032 and data record M 2034, data record Z 2032 is
inferred hierarchically linked to data records L and N 2036, 2038
in the same way as data record M 2034 is hierarchically linked to
data records L and N 2036, 2038.
[0114] The methodology discussed above with respect to FIG. 19 may
be clarified further with respect to FIGS. 21 and 22. In FIG. 21
incoming data record "t" 2130 is compared to existing data records
in existing data hierarchies 2110 and 2120 (e.g. data records
associated with the nodes of data hierarchies 2110 and 2120).
Suppose that data record "t" 2130 does not match any of these data
records. In this case data records "t" 2130 may be associated with
its own entity 2140 which is designated as a root node of data
hierarchy 2150 separate from existing data hierarchies 2110 and
2120 (e.g. data record "t" may be associated with a new entity
identifier and a self-identity link is formed with this entity
identifier).
[0115] If, however, data record "t" 2130 does match an existing
data record it may be integrated into an existing data hierarchy.
This scenario is depicted in FIG. 22. If data record "t" 2130
matches a data record in existing data hierarchies 2110, 2120 then
data record "t" may be linked with the entity with which the
matching data record is associated. If the match is a Soft Link
this link may be designated for review by an operator or user as
discussed above. Furthermore, in one embodiment, if the data record
to which data record "t" matches is not designated as primary (e.g.
did not originate from a data source designated as primary) the
link may also be designated for review. Here, data record "t" 2130
has matched to data record 2160 of node 2170 in existing data
hierarchy 2110. Thus, data record "t" will be associated with the
entity corresponding to node 2160 as well (e.g., a link formed
identically linking data record "t" and data record 2130).
[0116] Suppose now a data record comes in from another source. This
new data record may likewise be compared against data records in
existing data hierarchies. Two examples for such scenarios are
depicted in FIGS. 23 and 24. In FIG. 23 incoming data record "u"
2330 is compared to existing data records in existing data
hierarchies 2110, 2120 and 2310 (comprising node 2340 associated
with data record "t" 2130). If data record "u" 2330 matches data
record "t" 2130 data record "u" 2330 may be associated with entity
2340 (comprising data record "t" 2130) (either as a Hard_Link or
Soft_Link).
[0117] Suppose now, referring to FIG. 24, that incoming data record
"u" 2330 matches data record "t" 2130, but that data record "t"
2130 has matched to data record 2160 of node 2170 in existing data
hierarchy 2110 (as depicted with respect to FIG. 23). In this case,
data record "u" 2330 will be linked with node 2170 as well (e.g., a
link formed identically linking data record "u" 2330 and data
record "t" 2130). Again, if the match between data record "u" 2330
and data record "t" 2130 is a Soft Link the link may be tagged or
designated for review while if the match is a Hard_Link no such
review may be necessary.
[0118] While the above illustrations may be helpful, in many cases
an incoming data record may match multiple data records in one or
more existing data hierarchies. One example of a scenario of this
type is depicted with respect to FIG. 25. Generally, if an incoming
data record matches multiple existing data records the incoming
data records may be linked with the matching data record with the
highest score and the link may or may not be designated for review
by a user. Suppose, however, that an incoming data record matches
multiple existing data records in one or more existing data
hierarchies and the match score between the incoming data record
and each of the matching data records is identical. For example,
incoming data record "t" 2510 matches data record 2520 associated
with node 2522 of data hierarchy 2530 with a 7.1 match score, data
record 2524 associated with node 2526 of data hierarchy 2530 with a
7.1 match score and data record 2528 associated with node 2560 of
data hierarchy 2570 with a 7.1 match score. In this case, the
incoming data record may be associated with the data record
associated with the lowest number entity identifier. Continuing
with the above example, suppose node 2522 comprising data record
2520 corresponds to an entity having an entity identifier of "104",
node 2526 comprising data record 2524 corresponds to an entity
having an entity identifier of "108" and node 2560 comprising data
record 2528 corresponds to an entity having an entity identifier of
"110". In this case, data record "t" 2510 may be associated with
data record 2520 of node 2522 and thus associated with entity
identifier "104" (e.g. an identity link may be formed between
incoming data record "t" 2510 and data record 2520, for example
"104.1X:T, Y" where data record 2520 is "Y", such that incoming
data record "t" 2510 is associated with the same entity as data
record 2520, the entity in turn corresponding to node 2522).
[0119] Tasks may also be created such that this link is reviewed,
and may indicate whether the match score is lower than the Hard
Link threshold (but above the Soft Link threshold), that the
incoming data record has matched data records in multiple data
hierarchies, etc. For example, if the match score (e.g., between
the incoming data record and the multiple matching data records) is
above the Hard Link threshold, a review task may be created for a
user which indicates this, along with whether the multiple matching
data records lie in a single existing data hierarchy or multiple
existing data hierarchies. These review tasks may allow a user to
not only review the link that was created by MEI system 30, but the
other matching data records as well (e.g., matching but unlinked
data records) such that the user can determine if the created link
is correct and make any desired adjustments.
[0120] In addition to the above scenarios, it may also occur that
multiple incoming data records from multiple data sources may match
data records corresponding to a node in a data hierarchy. In this
case, in one embodiment, the incoming data record with the highest
match score from each data source may be linked to its respective
matching data record. An example of this scenario is depicted in
FIG. 26, where data record "t" 2610 and data record "u" 2620 may be
from one data source and data record "t" 2610 matches data record
2630 associated with node 2640 of data hierarchy 2650 while data
record "u" 2620 may also match data record 2630 of node 2640 of
data hierarchy 2650. Data record "v" 2670 and data record "w" 2680
may be from another data source and data record "v" 2670 matches
data record 2630 associated with node 2640 of data hierarchy 2650
while data record "w" 2680 matches data record 2630 of node 2640 of
data hierarchy 2650.
[0121] In one embodiment, if two data records from a data source
match an existing data record the data record with the highest
match score is linked to the existing data record. Thus, in the
example depicted, as data record `t` matches data record 2030 with
a score of 9.8, and data record "u" 2620 from the same source only
matches with a score 6.4, data record "t" 2610 is linked to data
record 2630. Similarly as data record "v" 2670 matches data record
2030 with a score of 7.3, and data record "w" 2680 from the same
source matches with a higher score of 8.6, data record "w" 2680 is
linked to data record 2630 (e.g., an identity link is formed
between data records 2610, 2630, and 2680 such that all these data
records are associated with one another and node 2640 of data
hierarchy 2650). In one embodiment, tasks may also be created such
that one or more of these links is reviewed by a user depending on
if the match score was above a certain level, or the data records
which had scores above a certain threshold (e.g. Soft Link or Hard
Link threshold) but which were not linked may be reviewed. Again,
as mentioned above, these tasks may indicate if the scores of the
unlinked data records were above a certain threshold, etc. In one
embodiment, if the lower (soft link) threshold is less than or
equal to 6.4, then all incoming records (independent of source)
with a match score of 6.4 or above may be linked to data record
2630 and a task may be created based on their pairwise scores
respective to the higher (hard link) threshold. More specifically,
2 or more data records from a given source may be allowed to
co-exist at the same node in a hierarchy tree assuming if the match
score for the data record indicates it is a best match (or meet the
tie break criteria of lowest unique identifier) and the match score
for the data record is equal to or greater than the lower
threshold. The actions to take with respect to this scenario, or
almost any other scenario imaginable, may be configurable by a user
of the system.
[0122] The above depictions of the operation of various embodiments
of the present invention may be useful when matching incoming data
records to existing data hierarchies, many times, however, sets of
data records may be received from information sources 34, 26, 28
where these incoming data records are arranged in an existing data
hierarchy (e.g., a set of incoming data records are hierarchically
linked, where these hierarchy links may be provided by information
sources 34, 26, 28). Thus, it may be desirable not only to
associate incoming data records with existing data records or
hierarchies, but to index incoming data hierarchies as well, in
other words, to associate incoming data records of an incoming data
hierarchy with existing entities and reconcile or composite the
incoming data hierarchy with any existing data hierarchies to which
the existing data records belong.
[0123] As may be imagined, reconciling data hierarchies may present
a variety of different problems. The most problematic of these
obstacles, however, may be the linking of the various data records
within each of the data hierarchies to composite (e.g., merge or
graft) data hierarchies based upon the matching of data records
within each of the hierarchies. For example, it may be relatively
simple to map an incoming data hierarchy to an existing data
hierarchy if every data record in the incoming hierarchy matches
only a single data record of an existing data hierarchy and the
hierarchy links between data records of the incoming hierarchy
mirror the hierarchy links between the corresponding matching data
records in the incoming data hierarchy. It may be more difficult,
however, when only a limited number (e.g., less than all) the data
records in an incoming data hierarchy match data records within an
existing data hierarchy, when data records in incoming hierarchies
match multiple data records in multiple existing data hierarchies,
when data records match between an existing data hierarchy and an
incoming data hierarchy but the hierarchy links between the two
data hierarchies do not correspond, etc. In cases such as these, it
may still desirable to reconcile an incoming data hierarchy with an
existing data hierarchy despite the occurrence of discrepancies
(e.g. mismatched data records in the data hierarchies, etc.).
[0124] Referring now to FIGS. 27A and 27B, a flow diagram for one
embodiment of a method for a match operation between data
hierarchies from various sources. These sources may be existing
data hierarchies, such as ones already existing in conjunction with
MEI, or may be received from information sources 34, 36, 38 which
may comprise reference sources such as Dun & Bradstreet,
Experian, Axciom, InfoUSA, etc. Thus, each of the data hierarchies
being compared may be from one or more of these sources and, via a
precedence definition of the sources themselves, each of the data
hierarchies may be designated as a primary source, a secondary
source, a tertiary source, etc. For example, if an existing data
hierarchy in the MEI is ranked higher than a data hierarchy from
InfoUSA, the existing data hierarchy may be designated as primary
while the data hierarchy from InfoUSA may be designated as a
secondary (or non-primary) source.
[0125] Specifically with reference now to FIG. 27A, for each
incoming data records of a data hierarchy received at step 2740,
this data record may be compared to existing data records at step
2742 (e.g. according to the method discussed with respect to FIG.
19). It can then be determined at step 2744 if the associated score
for the best matching candidate data record (e.g. best match score)
is equal or greater to the review threshold (e.g. above the Soft
Link threshold). If the best match score is above the review
threshold and if there is not a tie at the highest score at step
2746 (e.g. multiple candidate data records have the best match
score), the incoming data records may be identically linked with
the candidate records with the best match score at step 2748.
[0126] If, however, at step 2746 there are multiple candidate data
records with associated match scores equal to the best match score,
a top most parent for each of these candidate data records can be
determined at step 2750. In one embodiment, the top most parent for
a candidate data record may be a root node of a data hierarchy to
which the candidate data record belongs. The incoming data record
may then be identically linked with the candidate data record with
the best match score associated with the lowest entity identifier
and record made that the link corresponds to a single tree (e.g.
because there was only a single top most parent) at step 2754, or
identically linked with the candidate data record with the best
match score associated with the lowest entity identifier and record
made that the link corresponds to multiple trees at step 2756.
[0127] It can then be determined at step 2758 if the best match
score associated with a candidate data record is above or equal to
an automatic link (e.g. Hard Link) threshold. It is noted if the
best match score is below the threshold at step 2762 or above or
equal to the threshold at step 2760. At step 2764, then, it can be
determined if a task should be created, and if so, a task created
at step 2766. These tasks may allow for a user or operator to
manually review a link (either identical or hierarchy) created
between data records (and possibly the resulting compositing of
data hierarchies that result from the linking or comparison of data
records in various data hierarchies). In one embodiment, step 2766
the link may comprise a suffix denoting ambiguity in the linking
decision relating to the number of data hierarchies to which a data
record has been linked and a prefix denoting ambiguity in the
scoring. Thus, the step may be two-fold. First, the "suffix" (i.e.,
STree or MTree) may be determined indicating ambiguity in the
linkage decision, then g the "prefix" (i.e., HardLink or SoftLink)
ambiguity may be determined. The concatenation of the prefix plus
the suffix may then comprise the task type
[0128] In one embodiment, tasks may be created base upon various
determinations made during the matching or linking of data records.
For example, whether there were multiple candidate data records
with the best matching score at step 2746, whether there is one or
more top parent was identified at step 2752 (e.g. single tree or
multiple tree link), whether the best match score is above the
automatic link threshold determined at step 2758 (e.g. Hard Link or
Soft Link), etc. A task may be created, for example, if the link
between two data records falls below the automatic link threshold
(e.g. is a Soft Link) or more than one candidate data records has
the best match score.
[0129] Returning to step 2744, if the best match score is below the
review threshold (e.g. below the Soft Link threshold) it can be
determined at step 2768 if the incoming data record has a parent
(e.g. is associated with a node of a data hierarchy that is
parented by another node). If the incoming data record has a parent
then it may be hierarchically linked (P link type) with this parent
at step 2770 after which processing for this incoming data record
is complete at step 2772. If the incoming data record does not have
a parent it may be determined if the incoming data record has
children at step 2776 and, if so, this incoming data record
designated for post processing at step 2774, completing processing
for this data records at step 2772.
[0130] Moving now to FIG. 27B, one embodiment of a method for post
processing according to an embodiment of a match operation between
data hierarchies from various sources is depicted. For each of the
incoming data records designated for post processing at step 2774
(e.g., incoming data records not previously linked with a candidate
data record which are associated with a node which is a parent of
other nodes) it can be determined if any of the children of this
incoming data record have been linked to any existing data records
at step 2778. If so, and if more than one child of the parent has
been matched, as determined at step 2782, each of the highest
scoring children may be identified at step 2784. If there is more
than one child matching with the highest score, as determined at
step 2786, a hierarchy link may be formed between the top most
parent associated with the lowest entity identifier and the
incoming data record such that the top most parent of the data
hierarchy is designated as the parent of the incoming data record
at step 2788. Alternatively, at step 2788, it may be determined
which of the children with the highest score is associated with the
lowest identifier and the parent of this child may then be
hierarchically linked to the parent of the hierarchy to which the
child matched was. If, however, at step 2786 it is determined that
there is only one matching child with the highest score a hierarchy
link may be formed between the top most parent associated with the
lowest entity identifier and the incoming data record such that the
top most parent of the data hierarchy is designated as the parent
of the incoming data record at step 2790. A task can then be
established such that a user can review this hierarchy link(s) at
step 2792 whereupon processing for the particular incoming data
record/hierarchy is complete at step 2794.
[0131] It may be useful to depict various examples of the
application of embodiments of the present invention. One such
example is depicted in FIG. 28, depicting an example of one
embodiment of compositing data hierarchies. In particular, incoming
data hierarchy 2810 may comprise data record "m" 2812, data record
"n" 2814 and data record "l" 2816, where data records 2812, 2814
and 2816 are hierarchically ordered, specifically where data record
"l" 2816 is the parent of both data record "m" 2812 and data record
"n" 2814 (e.g. data record "l" 2816 is hierarchically linked to
data record "m" 2812 such that data record "m" 2812 is parented by
data record "l" 2816).
[0132] Data record "n" 2814 is compared against data records
associated with data hierarchies 2830, 2840 (e.g. data records of
entities associated with nodes of the data hierarchies). Here, data
record "n" 2814 may match data record 2822 of node 2824 of data
hierarchy 2830. Thus, data record may be identity linked to data
record 2822 of data hierarchy 2830. Notice however, that data
record "l" 2816 is both unmatched and is a top most parent of a
child which has been matched (e.g. data record "n" 2814), thus data
records 2856 and 2858 comprising node 2860, the top most parent of
data hierarchy 2730 to which data record 2722 (e.g. matching data
record "n" 2814) may be identified and a hierarchy link formed
between data record "l" 2816 and data record 2856 such that data
record "l" 2816 is parented by node 2860 (e.g. entity corresponding
to node 2860 parents data record "l" 2816). Similarly, data record
"m" 2812 is unmatched. However, as data record "m" 2812 has a
parent (e.g. data record "l" 2816) a hierarchy link between data
record "l" 2816 and data record "m" 2812 may be established (or
remain) such that data record "m" 2812 remains parented by data
record "l" 2816 despite the fact that data record "l" 2816 is now
parented by node 2860 of data hierarchy 2830.
[0133] In this manner, data hierarchy 2810 is composited with data
hierarchy 2830. In this case, a task may be created such that the
user can review the compositing of data hierarchies 2810 and 2830
(e.g., the links formed between data records to accomplish the
compositing of the data hierarchies) or the matching of data record
"n" 2814 to data record 2822. This task may identify whether the
match score between data record "n" 2714 and data record 2822 was
above the review threshold or above an automatic link
threshold.
[0134] Turning to FIG. 29, another scenario for an example
compositing of data hierarchies is depicted. In particular,
incoming data hierarchy 2910 may comprise data record "m" 2912 and
data record "l" 2916, where data records are hierarchically ordered
such that data record "l" 2916 is the parent of data record "m"
2912. Data record "m" 2912 and data record "l" 2916 are compared
against data records associated with data hierarchies 2930 and 2940
(e.g. data records of entities associated with nodes of the data
hierarchies). Here, data record "l" 2916 may match data record 2922
of entity 2924 of data hierarchy 2930. Thus, data record "l" 2916
may be identity linked to data record 2922 of data hierarchy 2930.
Notice however, that data record "m" 2912 is unmatched. As data
record "m" 2912 is parented by data record "l" 2916, however, a
hierarchy link where data record "m" 2912 is parented by data
record "l" 2916 may be established or maintained despite the fact
that data record "l" 2916 is now linked to data record 2922 of node
2924 of data hierarchy 2930. In this manner, data hierarchy 2910 is
composited with data hierarchy 2930. Again, a task may be created
such that the user can review the compositing of data hierarchies
2910 and 2930 or the matching of data record "l" 2916 to data
record 2922. This task may be based on whether the match score
between data record "l" 2916 to data record 2922 was above the
review threshold or above an automatic link threshold.
[0135] In the foregoing specification, the invention has been
described with reference to specific embodiments. However, one of
ordinary skill in the art appreciates that various modifications
and changes can be made without departing from the scope of the
invention as set forth in the claims below. Accordingly, the
specification and figures are to be regarded in an illustrative
rather than a restrictive sense, and all such modifications are
intended to be included within the scope of invention.
[0136] Benefits, other advantages, and solutions to problems have
been described above with regard to specific embodiments. However,
the benefits, advantages, solutions to problems, and any
component(s) that may cause any benefit, advantage, or solution to
occur or become more pronounced are not to be construed as a
critical, required, or essential feature or component of any or all
the claims.
* * * * *